The ERP Foundation Is the Whole Game. AI Changes How Fast You Can Win.

Karl-Anthony Towns just won a championship with the Knicks. When asked what it actually takes, Towns did not talk about the games. He talked about the hours nobody sees.

As a kid, he was in the gym about five hours a day. His father coached high school basketball, so Towns practiced with the freshman team, then the JV, then the varsity, with his dad rebounding for him the whole time. He calls that early access an advantage. Not his height. Not raw talent. Access, and the discipline to use it before anyone told him he had to.

It still runs his career. In season, team practice does not start until 11. Towns is at the facility by 7, four hours early, working in his words "when it's dark and no one's there," until about 1:30.[1] The championship is what the world sees. The 7 AM work is what makes it possible.

A successful ERP works in much the same way. The foundation is the part nobody sees, and it decides everything built on top of it. It is the whole game.

The foundation nobody gets excited about

Nobody gets excited about an ERP. It does not trend. It does not launch with a keynote. But it is the foundation the whole business runs on, and when a company treats it as optional, the reports it produces cannot be trusted. Bad data is not free. Gartner puts the average cost of poor data quality at $12.9 million a year.[2]

Where the time actually goes

It looks tidy on paper. In practice this is where projects stall, and it is worth understanding why.

Master data cleanup means reconciling years of duplicates and dead records by hand, department by department. Process alignment means deciding which of your habits to keep and which to drop. For most companies the right move is to adopt the system's best practices rather than bend the system around old habits, but not always. A jewelry business cannot run standard replenishment on a one-of-a-kind piece, and someone has to know which exceptions are real and which are just history. Documentation and training get deferred for a defensible reason, since anything you write goes stale the moment the configuration changes, so it waits until go-live and then gets cut for time.

None of this work is exciting. All of it decides whether the system works. And until recently, all of it ran at human speed, which is why ERP projects have always been slow and expensive.

What AI changes

AI is good at exactly this work. It profiles and cleans data at scale, holds the standards as new data arrives, learns the patterns of your industry, and keeps the documentation current instead of stale. On the judgment calls it helps too, drafting the exception a business like jewelry needs and flagging the conflicts and edge cases that otherwise surface late and break go-live.

It does not replace the people who know your business. It does the heavy lifting so they can spend their time on the calls that require judgment. The foundation gets built faster, kept cleaner, and delivered at lower cost and lower risk. Human judgment still makes the calls.

And the payoff is real. A clean foundation turns the month-end close from a week-long scramble into a few days.[3] It makes margin visible by product, customer, and channel as it happens. And paired with AI forecasting, it changes the demand picture. McKinsey found that AI-driven forecasting can cut forecast error by 20 to 50 percent and reduce lost sales from stockouts by up to 65 percent.[4]

Here is the whole foundation in one view. Each piece at human speed, each piece with AI, and what a clean build puts on the board.

The part most companies miss

The companies that win the next few years in ERP will not be the ones that skip the foundation. They will be the ones that build it faster, and keep it clean.

ERP has long been the most change-resistant corner of the enterprise, and for understandable reasons. The complexity was real and the foundation work was manual, which is why implementation has always been expensive and slow. That is exactly what is shifting. The teams that use AI to compress that work, without cutting corners on quality, will change the economics of the whole category. They will deliver cleaner implementations, lower risk, and a shorter path to value, and they will make the cost of an ERP look very different than it does today. The vendors building AI in from the foundation up hold a real and durable advantage.

That makes waiting the risky position, not the safe one. The teams that move early reset what the work costs. Push it a few exits down the roadmap and you may not get the chance to catch up.

Towns did not show up at 7 AM because it was fun. He showed up because the work nobody sees is what decides the outcome. A successful ERP is the same. The difference now is that the 7 AM work no longer has to take as long.

Notes

1. Quotes from Karl-Anthony Towns are from The Howard Stern Show, SiriusXM, June 16, 2026.

2. Cost of poor data quality: Gartner. https://www.gartner.com/en/data-analytics/topics/data-quality

3. Median month-end close and the reduction from automation and a continuous close: APQC and Ventana Research. https://www.apqc.org/resource-library/resource/cycle-time-perform-monthly-close

4. Forecast error and lost sales reductions: McKinsey, "AI-driven operations forecasting in data-light environments," February 15, 2022. https://www.mckinsey.com/capabilities/operations/our-insights/ai-driven-operations-forecasting-in-data-light-environments